Neither of the two routes already sketched leads directly into middle or northern Europe. The trend in both is toward the Mediterranean. We must now consider the third and last route, which is of chief interest to us. We have already seen that the Black Sea prevented all migrations northward from Asia Minor except at the Hellespont. Eastward from the Black Sea lies the Caspian, probably much larger in glacial times. The two seas are separated by the forbidding, almost unbroken, mountain barrier of the Caucasus; but a narrow passage at each end is left. East of the Caspian Sea must lie the point where a more northerly westward route diverges from the road through Asia Minor. Our third route starts, therefore, from the northern edge of the Iranian plateau, perhaps mostly from Turkestan, and runs westward north of the great barrier of seas and mountains just described. It follows the great steppe or prairie which stretches through southern Siberia and Russia into Hungary. Its western portion lies along the valley of the lower Danube, the great east and west artery of communication and migration through Europe. It lies farther north than any other great route, and leads over steppe instead of through forest. As the Arabia-Africa route was the first to be traversed, this may well have been the last. Furthermore, the route through Asia Minor, ending in a sort of _cul de sac_, may easily have become well inhabited and hence less open before the Neolithic period had begun in northern Europe. It was by no means the most attractive route. It offered far less to people in the collecting stage than the well-watered parklands of Asia Minor. The steppe offers to the hunter few means of concealment or approach to the game. The animals are swift and wary. In any migration of peoples toward the frontier, the hunters lead the advance and spread out like an army of scouts. Every river which crossed the steppe would offer to them a tempting by-road leading off into the forests of Siberia or Russia. How deeply they would penetrate into the primeval forest or away from the river valleys is still a question. Very likely they would find their best hunting-grounds not very far from the northern edge of the steppe, where the forest is less dense. This question we cannot yet answer. But most of European Russia is well watered, and here these hunters would find themselves at home. The main route of the steppe would be left for a very different population. The piedmont zone of grasslands in Turkestan was an ideal land for primitive agriculturists practising a hoe-culture, as at Anau. The northern edge of this steppe zone, where it joined the forest, may have been equally favorable.